Please use this identifier to cite or link to this item: http://repositorio.ufla.br/jspui/handle/1/46246
Title: Dados de área na família GAMLSS em estudos epidemiológicos
Other Titles: Area data in the gamlss family in epidemiological studies
Authors: Oliveira, Izabela Regina Cardoso de
Scalon, João Domingos
Bastiani, Fernanda de
Keywords: Estatística espacial
Modelos de regressão flexíveis
Saúde pública
Tuberculose bovina
Dengue
Flexible regression models
Public health
Spatial statistics
Bovine tuberculosis
Issue Date: 10-May-2021
Publisher: Universidade Federal de Lavras
Citation: BRITO, A. de L. Dados de área na família GAMLSS em estudos epidemiológicos. 2021. 91 p. Dissertação (Mestrado em Estatística e Experimentação Agropecuária) – Universidade Federal de Lavras, Lavras, 2021.
Abstract: The progress in the field of statistical analysis has been increasingly significant in recent years. In particular, regression models comprise a set of statistical tools that have received major contributions in a short period of time. These models are one of the most used tools in the scientific world to describe several phenomena in the most varied areas of knowledge. Since the beginning, the normal linear model has been used in many scientific researches for data modeling. However, due to the limitations found in this model, the Generalized Linear Models were introduced, which encompass more probabilistic distributions for the response variable. Subsequently, Generalized Additive Models were proposed, which made the relationship between variables more flexible. Then the GAMLSS models were introduced, which can be seen as a generalization of the other models mentioned above. This new class of models allows not only a greater number of probabilistic distributions for modeling the response variable, but also greater flexibility for the relationship between the variables, as well as the modeling of other distribution parameters, in addition to the location. Recently, an adaptation was made in the GAMLSS models to incorporate the effect of a spatial dependence structure when the assumption of independence of the observations of the response variable is not met and there is correlation in space. In this study, we aimed to study the GAMLSS models in the context of spatial analysis and apply them to real data. For this, data on the occurrence of bovine tuberculosis in the state of Minas Gerais and dengue in the state of Paraíba were used. Satisfactory adjustments were obtained for both databases, even when they presented structural problems such as strong asymmetry and kurtosis and a problem of overdispersion. For dengue data, a spatial component was introduced in the model through an intrinsic autoregressive model, since the data showed significant spatial autocorrelation.
URI: http://repositorio.ufla.br/jspui/handle/1/46246
Appears in Collections:Estatística e Experimentação Agropecuária - Mestrado (Dissertações)

Files in This Item:
File Description SizeFormat 
DISSERTAÇÃO_Dados de área na família GAMLSS em estudos epidemiológicos.pdf14,19 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.